Body size trends in Neogene tortoises

30.05.2017

Test paleoTS with Fossil Checklist data (but is probably of no use, because they report average body sizes (means, median, something else? what are the respective sample size? maybe ask the authors!?), so this is just for playing around).

Raw data:

library(paleoTS)
#setwd("//naturkundemuseum-berlin.de/MuseumDFSRoot/Benutzer/Julia.Joos/Eigene Dateien/MA")
test<-read.csv("test26.5.csv", sep=";", header=TRUE)
test

The first plot shows mean Cl size for each taxon as a single data point, so each data point is one species (in this case this equals one individual, since I don’t have sample sizes), even within time bins.

Test1 <- test %>%
  mutate(mm = CL_mean, vv=0, nn= n, tt=Age_mean) %>%
  dplyr::select(mm, vv, nn, tt)
paleoTest1 <-as.paleoTS(Test1$mm, Test1$vv, Test1$nn, Test1$tt, MM = NULL,
                        genpars = NULL, label = "Testudinidae body size evolution mode")
paleoTest1

$mm
 [1] 107.5  24.0  28.0  23.0 165.0  80.0 120.0 175.0  11.0  90.0  33.0 150.0 115.0 125.0  47.0  75.0  37.5 192.5 195.0   9.0
[21]  83.0  95.0  40.0  58.0  34.0  29.0  20.0  23.0  45.0  60.0  50.0 100.0 190.0  22.0 100.0  90.0 195.0 120.0  20.0  12.0
[41]  22.5  15.0 186.0  62.5  87.5  85.0  50.0  20.0  25.0  13.0  27.5  52.0 122.5  80.0  37.5 150.0  79.0  98.0  96.0  88.0
[61]  90.0 100.0  46.0 100.0 120.0 110.0  40.0 150.0  90.0 120.0  27.5

$vv
 [1] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
[63] 0 0 0 0 0 0 0 0 0

$nn
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[63] 1 1 1 1 1 1 1 1 1

$tt
 [1]  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000  0.00000 -1.28765
[14] -1.28765 -1.28765 -1.28765 -1.28765  4.99550  3.23250  1.40950 -1.22465 -1.22465 -1.22465 -1.22465 -1.22465 -1.22465
[27] -1.22465 -1.23000 -1.23000 -1.23000 -1.23050  0.56950  0.56950  0.56950  0.56950  0.56950  0.56950  0.56950  0.56950
[40]  0.56950  0.56950  0.56950  0.56950  0.56950 -0.84000 -0.84000 -0.84000 -0.84000 -0.84000 -0.84000 -0.84000 -0.84000
[53] -0.89715 -0.89715 -0.89715 -0.89715 -1.29350 -1.29350 -1.29350 -1.29350 -1.29350 -1.29350 -1.29350 -1.29350 -1.29350
[66] -1.29350  0.00635  0.00635  0.05050  1.00650  1.00650

$MM
NULL

$genpars
NULL

$label
[1] "Testudinidae body size evolution mode"

$start.age
[1] 1.2935

$timeDir
[1] "increasing"

attr(,"class")
[1] "paleoTS"

plot(paleoTest1)

This is the underlying data for Test1:

Test1

For the second plot, I averaged CL means across taxa for each time bin, which leaves one data point per time bin, comprising all taxa within the respective bin:

Test2 <- test %>%
  group_by(Age_mean) %>%
  summarise(mm = mean(CL_mean), nn=n(), vv=var(CL_mean)) %>%
  mutate(tt=Age_mean) %>%
  dplyr::select(mm, vv, nn, tt)
# NA: column 2, rows 3, 10, 13, 14, 15
Test2[3,2] <- 0
Test2[10,2] <- 0
Test2[13,2] <- 0
Test2[14,2] <- 0
Test2[15,2] <- 0
paleoTest2 <-as.paleoTS(Test2$mm, Test2$vv, Test2$nn, Test2$tt, MM = NULL,
                        genpars = NULL, label = "Testudinidae body size evolution mode")
paleoTest2

$mm
 [1]  92.70000  79.90000  50.00000  42.66667  51.28571  97.50000  45.00000  83.87500  95.00000  90.00000  87.30769  73.75000
[13]   9.00000 195.00000 192.50000

$vv
 [1]  398.6778 1542.5500    0.0000  346.3333  810.5714 2429.1667  833.6429 3589.5511 6050.0000    0.0000 4816.0224 4278.1250
[13]    0.0000    0.0000    0.0000

$nn
 [1] 10  5  1  3  7  4  8 12  2  1 13  2  1  1  1

$tt
 [1] 0.00000 0.00585 0.06300 0.06350 0.06885 0.39635 0.45350 1.29350 1.29985 1.34400 1.86300 2.30000 2.70300 4.52600 6.28900

$MM
NULL

$genpars
NULL

$label
[1] "Testudinidae body size evolution mode"

$start.age
NULL

$timeDir
[1] "increasing"

attr(,"class")
[1] "paleoTS"

plot(paleoTest2)

Since “real” variances and sample sizes are available when pooling all taxa, you can even fit models (as you should be able to in the end). (when I remember correctly, the model with the highest Akaike.wt is the best supported one, in this case this would be URW = random walk)

a=fit3models(paleoTest2, silent=FALSE, method="AD", pool=FALSE)   #not working with Test1, because no variances/sample sizes available, I guess

Comparing 3 models [n = 14, method = AD]

            logL K     AICc Akaike.wt
GRW    -70.40398 2 145.8989     0.373
URW    -71.26818 1 144.8697     0.625
Stasis -75.70460 2 156.5001     0.002

str(a)

'data.frame':   3 obs. of  4 variables:
 $ logL     : num  -70.4 -71.3 -75.7
 $ K        : num  2 1 2
 $ AICc     : num  146 145 157
 $ Akaike.wt: num  0.373 0.625 0.002

a$AICc[1] # not sure what this tells me...

[1] 145.8989

This is the underlying data for Test2:

Test2

TO DO:

06.06.2017

Try paleoTS with some first real data. Here is the underlying data:

tidyCL<-read.csv("tortoises_tidy.csv", sep=";", header=TRUE)
tidyCL

Prepare data for conversion to paleoTS-object:

TidyCL <- tidyCL %>%
  select(MAmin, Mamax, CL) %>%
  filter(CL != "NA") %>%
  mutate(tt= (MAmin+Mamax)/2) %>% # create mean age
  group_by(tt) %>% #create time bins
  summarise(mm=mean(CL), vv=var(CL), nn=n()) #create means etc. for each time bin 
TidyCL[is.na(TidyCL)]<-0 #subset NAs with O for 
TidyCL

bins <- tidyCL %>%
#  select(MAmin, Mamax, CL) %>%
  filter(CL != "NA") %>%
  mutate(tt= (MAmin+Mamax)/2) %>% # create mean age
  group_by(tt)
bins

library(paleoTS)
paleoTidyCL <-as.paleoTS(TidyCL$mm, TidyCL$vv, TidyCL$nn, TidyCL$tt, MM = NULL, genpars = NULL, label = "Testudinidae body size evolution mode")
paleoTidyCL

$mm
 [1]  195.0000  180.5000  860.3333  880.0000 1200.0000  278.0000 1200.0000  500.0000  500.0000  190.7500  370.0000  400.0000
[13]  400.0000

$vv
 [1]      0.0000     40.5000 307760.3333      0.0000      0.0000      0.0000      0.0000      0.0000      0.0000    911.9286
[11]      0.0000      0.0000      0.0000

$nn
 [1] 1 2 3 1 1 1 1 2 1 8 1 1 1

$tt
 [1]  0.000  1.230  2.180  2.730  3.730  6.730  7.730  8.330 10.080 11.230 16.730 21.345 32.180

$MM
NULL

$genpars
NULL

$label
[1] "Testudinidae body size evolution mode"

$start.age
[1] 1.77

$timeDir
[1] "increasing"

attr(,"class")
[1] "paleoTS"

plot(paleoTidyCL)

fit3models(paleoTidyCL, silent=FALSE, method="AD", pool=FALSE)   #not working with Test1, because no variances/sample sizes available, I guess

Comparing 3 models [n = 12, method = AD]

             logL K     AICc Akaike.wt
GRW     -94.17833 2 193.6900     0.001
URW    -104.38851 1 211.1770     0.000
Stasis  -87.43929 2 180.2119     0.999

Map <- tidyCL %>%
  select(Genus, Taxon, Latitude, Longitude, Country, CL, PL) %>%
  group_by(Latitude) %>%
  mutate(count= n())
mapWorld <- borders("world", colour="azure3", fill="azure3") # create a layer of borders
mp <- Map %>%
  ggplot(aes(Longitude, Latitude)) + mapWorld +
#  geom_point(fill="red", colour="red", size=0.5) +
  geom_point(aes(Longitude, Latitude,colour=CL, size=count))
mp

library(plotly)
ggplotly(mp)

We recommend that you use the dev version of ggplot2 with `ggplotly()`
Install it with: `devtools::install_github('hadley/ggplot2')`

TO DO:

• map localities with differing colors for: CL available, CL extrapolated (from PL or figures), CL missing
• complete data set!
• get missing refences/make list of missing references

08.06.17

Map all localities with sample size and age indicated (regardless of whether CL information is available):

test<-read.csv("tortoises13-04.csv", sep=";", header=TRUE)
colnames(test)[6] <- "Mamin"
colnames(test)[7] <- "Mamax"
Test <- test %>%
  select(Locality, Country, Latitude, Longitude, Mamin, Mamax, Epoch, Genus, Species, Taxon, CL) %>%
  mutate(Age= (Mamin+Mamax)/2) %>%   # create mean age
  group_by(Latitude) %>%
  mutate(count= n())
mapWorld <- borders("world", colour="azure3", fill="azure3") # create a layer of borders  
  
map <- Test %>%
  ggplot(aes(Longitude, Latitude)) + mapWorld +
  #geom_point(fill="red", colour="red", size=0.5) +
  geom_point(aes(Longitude, Latitude,colour=Age, size=count))
map

ggplotly(map)

We recommend that you use the dev version of ggplot2 with `ggplotly()`
Install it with: `devtools::install_github('hadley/ggplot2')`

This is an R Markdown Notebook. When you execute code within the notebook, the results appear beneath the code.

Try executing this chunk by clicking the Run button within the chunk or by placing your cursor inside it and pressing Ctrl+Shift+Enter.

Add a new chunk by clicking the Insert Chunk button on the toolbar or by pressing Ctrl+Alt+I.

When you save the notebook, an HTML file containing the code and output will be saved alongside it (click the Preview button or press Ctrl+Shift+K to preview the HTML file).